Improving Manufacturing Processes by Using Predictive Maintenance with Machine Learning and Lean Six Sigma Methods

Abstract

Unplanned stoppages and inefficient maintenance in industrial settings result in significant costs and reduced productivity. Old maintenance methods, like fixing things after they break or doing regular checks, are either expensive or not always dependable. This study uses machine learning to improve maintenance, combining predictive maintenance with lean six sigma. The AI4I 2020 dataset was used along with techniques like random forest, support vector machine, and artificial neural networks to predict when equipment might fail. The DMAIC method helped make maintenance more efficient. Because of this, the failure rate dropped from 3.39 to 2.00%, Overall equipment effectiveness improved, and the process became more stable, as shown by better Cp and Cpk values. However, challenges such as high implementation costs, poor data quality, and the need for employees to adapt to these new methods remain. Despite these challenges, predictive maintenance and lean six sigma show strong potential for improving industrial operations. Future research should focus on developing real-time predictive maintenance systems based on the Internet of Things (IoT) and explainable AI models to further automate and optimize industrial processes.